Freshwater planarians of the species D. japonica that have been maintained in the laboratory for over a decade were primarily used. To compare species differences, Dugesia dorotocephala and G. tigrina planarians (both listed as “brown planarians”, Carolina Biological Supply, Burlington, NC) were used in some bulk lethality experiments. All planarians were stored in 1x Instant Ocean (IO, Blacksburg, VA) in Tupperware containers and kept at 18°C in a Panasonic refrigerated incubator in the dark. The animals were fed organic beef liver (obtained from a local farm) or chicken liver (Bell and Evans, Fredericksburg, PA) once a week. The containers were cleaned twice a week following standard protocols (Dunkel et al, 2011). Fully regenerated worms starved ≥5 days and gliding normally in the container were used for all experiments. Planarians were selected to fall within a certain range of sizes, with larger planarians used for amputation/regeneration experiments. Thus, the final sizes of adult planarians and regenerating tails within a given experiment were similar (HTS: ∼3–6 mm, bulk: ∼5–8 mm, biochemical assays: ∼8–12 mm). To induce regeneration/development, intact planarians were amputated on day 1 via a transverse cut between the auricles and the pharynx with an ethanol-sterilized razor blade. Chemical exposure began within 3 h of amputation.

Pure glyphosate (CAS # 1071-83–6) was purchased from Sigma-Aldrich (St. Louis, MO) and had a purity ≥98%. Diquat dibromide (CAS # 6385-62–2, analytical standard) and Pelargonic acid (CAS # 112-05–0, purity 97.5%) were purchased from ChemService (West Chester, PA). Two formulations of Roundup^® were tested: Roundup^® Concentrate Plus (RC, product number LB5778, active ingredients: 18% glyphosate isopropyl amine salt, 0.73% diquat dibromide) and Roundup^® Ready-to-Use Weed and Grass Killer III (RR, product number LB5135, active ingredients: 2% glyphosate isopropyl amine salt, 2% pelargonic and other related fatty acids). The original commercial stocks were stored at room temperature in the dark. To compare equivalent concentrations of the free acid form of glyphosate to the salt forms found in the GBHs, the acid equivalent (ae) concentrations of glyphosate in each formulation were used. Thus, glyphosate and both Roundup^® formulations were compared at equivalent concentrations of glyphosate (ae, 1, 3.16, 10, 31.6, 100, 316 and 1,000 μM, Table 1). Diquat dibromide was tested at the equivalent concentrations found in the tested concentrations of RC, which corresponded to 0.0092, 0.029, 0.092, 0.29, 0.92, 2.9, and 9.2 mg/L. Pelargonic acid was tested at the equivalent concentrations found in the tested concentrations of RR, which corresponded to 0.282, 0.892, 2.82, 8.92, 28.2, 89.2, and 282 mg/L. In the HTS experiments, 100 μM L-ascorbic acid (CAS # 50-81–7, Sigma-Aldrich) was assayed as a negative control. Stocks of 10X the highest tested concentration of each test compound were prepared fresh in IO water on the day of experimental set-up.

To test whether lethality was affected with different exposure set-ups and for possible species differences in susceptibility (Ireland et al, 2020), D. japonica, G. tigrina and D. dorotocephala planarians were exposed in bulk to RR or RC at concentrations corresponding to 31.6, 100, or 316 µM glyphosate ae or to IO water (solvent control). These concentrations were chosen to cover sublethal and lethal concentrations seen in the HTS experiments with D. japonica. D. japonica planarians were also exposed to 100, 316, and 1,000 µM glyphosate and 1,000 µM RR or RC. Planarians were exposed in tissue culture-treated 12-well plates (Genesee Scientific). For D. japonica and G. tigrina experiments, 3 wells each containing 6 planarians in 1.2 mL of the test solution were tested. For D. dorotocephala experiments, 4 wells each containing 6 planarians in 1.2 mL of test solution were tested. Thus, in all tests the ratio of chemical/planarian (200 µL/planarian) was consistent with HTS. Lethality was manually scored at day 2, day 4, day 8, and day 10. The D. japonica experiments were also checked on day 7 and day 12 for comparison to HTS. To statistically evaluate whether exposure type (96-well or bulk exposure) had a significant effect on D. japonica lethality, a generalized loglinear model was created testing the interactions of lethality:concentration:worm type (adult or regenerating): exposure type. The statistical significance of each term was determined by dropping each term and testing the significance compared to the original model using a one-way ANOVA. p-values for these comparisons are provided in Supplementary Table S2. The same approach was used to evaluate the effect of species by testing the interactions of lethality:concentration:worm type (adult or regenerating):species. p-values for the species comparison are provided in Supplementary Table S3.

All tested formulations caused significant lethality to both adult and regenerating D. japonica planarians at 1 mM on both days 7 and 12 when exposed in 96-well plates. Significant lethality was observed at lower concentrations for the two GBHs than for glyphosate alone. Significant lethality in both adult and regenerating planarians was induced by 316 μM RR, while 316 µM RC only caused significant lethality in adult planarians (Figure 1). As previous studies with planarians have primarily used bulk exposure, we compared the lethal effects of glyphosate, RR and RC when exposing single planarians in 96-well plates or 6 planarians/well in 12 well plates (Figure 1). We found a slight increase in lethal effects when using 96-well plates, with the exception of glyphosate exposure in regenerating planarians, which caused greater lethality in bulk exposure versus 96-well plates at 1 mM. The significance of these observations were quantified by analyzing the interaction between lethality, concentration, worm type (adult or regenerating) and exposure type (96-well or bulk) using a generalized linear model. At both day 7 and day 12, exposure type and worm type had a significant effect on lethality for glyphosate and RC (Supplementary Table S2). Bulk exposure also showed greater variability across the independent wells/replicates (individual dots in Figure 1) than the 96-well plate data.

We also compared the lethal effects of RR and RC on two other planarian species, D. dorotocephala and G. tigrina, using bulk exposure in 12-well plates. D. dorotocephala and G. tigrina were more sensitive than D. japonica to both RR and RC as 316 µM of either GBH caused at least 50% lethality by day 10 in both developmental stages (adult and regenerating), though the temporal dynamics differed (Supplementary Figure S1). Analysis of the interaction between lethality, concentration, worm type (adult or regenerating) and species using a generalized linear model revealed that species and worm type had a significant effect on lethality for both RR and RC.

Morphological and behavioral defects at sublethal concentrations were assessed using an automated platform that evaluates morphology and various behaviors. Screening was done on both adult and regenerating planarians to assess whether differential sensitivity was present in the two developmental stages. Assessments were done on day 7 and day 12 to evaluate the dynamics of effects. A comparison of the LOEL for each formulation and endpoint is presented in Figure 2. Glyphosate did not induce any significant effects at sublethal concentrations. RR induced specific effects in stickiness at day 7 and in noxious stimuli strength at day 12 at 100 µM. Notably, these effects were only seen in adult planarians, whereas regenerating planarians showed no sublethal effects of RR exposure. RC exposure induced significant sublethal behavioral effects. Although 316 µM RC was not significantly lethal to regenerating planarians as it was in adult planarians, this concentration caused severe toxicity in the regenerating planarians. Regenerating planarians exposed to 316 µM RC showed abnormal body shapes including corkscrews, C-shapes, and contraction on both day 7 and day 12 (Figure 3A). On day 7 and day 12, 78% (n = 23; p = 7.4 × 10⁻⁹, Fisher’s Exact Test with BH correction) and 67% (n = 21; p = 7.0 × 10⁻⁷, Fisher’s Exact Test with BH correction) of regenerating planarians exposed to 316 µM RC showed any abnormal body shape, respectively. In-plate controls and all lower concentrations did not have any planarians with abnormal body shapes. These abnormal shapes were correlated with locomotion defects, including reduced speed (Figure 3B), increased resting, and defects in phototaxis and the noxious heat response. Together, these suggest that 316 µM RC caused systemic toxicity in regenerating planarians. Adult planarians showed greater sensitivity to RC exposure, with significant sublethal effects on motility/locomotion, manifested as decreased speed and increased resting, starting at 31.6 µM (Figure 3B).

We tested whether 12-day exposure to the highest tested sublethal concentration (100 µM) glyphosate, RR, or RC affected AChE activity in adult planarians (Figure 4A). Only the positive control diazinon showed a statistically significant decrease in AChE activity (one-way ANOVA, followed by Dunnett’s test, p<<0.01). None of the GBHs had a statistically significant effect on AChE activity.

Many studies have suggested that glyphosate and GBHs can cause oxidative stress (reviewed in (Costas-Ferreira et al, 2022)). Thus, we evaluated the effects of glyphosate/GBH exposure on the activity of a common biomarker of oxidative stress—GST—at the highest concentration of glyphosate (ae) that was non-lethal for all three compounds (i.e., 100 µM). GST has been previously characterized in D. japonica (Zhang et al, 2020). We found no significant effects using a one-way ANOVA on GST activity with any of the tested compounds at 100 µM in adult planarians after 12 days of exposure (Figure 4B).

RC and RR also contain other active ingredients besides glyphosate (diquat dibromide and pelargonic acid, respectively), that may contribute to the toxicity of the GBHs. Thus, we screened diquat dibromide and pelargonic acid alone at the equivalent concentrations that would be found in the tested concentrations of the respective GBHs. Pelargonic acid did not show any significant effects in any endpoints, suggesting no toxicity to planarians up to 282 mg/L. In contrast, diquat dibromide caused lethality at day 12 at the highest tested concentration (9.2 mg/L, equivalent to the concentration of diquat dibromide found in RC at 1,000 µM glyphosate ae) in adult planarians. Diquat dibromide also caused many sublethal behavioral effects across different endpoints, starting at 0.29 mg/L (Supplementary Figure S2A). For adult planarians, the overall LOEL for diquat dibromide was equivalent to the LOEL found for RC (RC at 31.6 µM glyphosate ae contains 0.29 mg/L diquat dibromide), though the toxicological profiles (exact endpoints affected) differed slightly. Regenerating planarians exposed to diquat also demonstrated abnormal behavior starting at 2.9 mg/L, though no significant lethality was induced up to 9.2 mg/L. As in RC, exposure to diquat induced abnormal body shapes. However, unlike RC, these were primarily found in adult planarians and were highly dynamic as the planarians writhed uncontrollably in C- and S-like shapes (Supplementary Figure S2B). In adult planarians, abnormal body shapes were induced starting at 0.92 mg/L on both day 7 and day 12, whereas regenerating planarians only demonstrated abnormal body shapes on day 12 at 9.2 mg/L (Supplementary Figure S2C).

We compared the toxicity of glyphosate alone to that of two common GBHs in freshwater planarians. Both GBHs were more toxic than pure glyphosate, which only showed effects in lethality at the highest test concentration (1 mM). Regarding lethality, RR was the most toxic, with lethal effects in both adult and regenerating planarians at 316 µM. RC was slightly less toxic than RR, as significant lethality was not observed at 316 µM RC in regenerating planarians. However, these planarians displayed obvious signs of toxicity (abnormal body shapes, immobility). RC had an overall LOEL of 31.6 µM in adult planarians and caused much stronger sublethal phenotypic effects than RR as effects on multiple locomotion-based endpoints were observed. In contrast, RR only caused increased stickiness and a more sensitive response to noxious heat (NS strength endpoint) only in adult planarians at 100 µM. We have previously found that inhibition of AChE activity can correlate with effects in these endpoints (Ireland et al, 2022). As an organophosphonate, glyphosate is a weak inhibitor of mammalian AChE (Larsen et al, 2016). However, neither glyphosate nor the GBHs had a statistically significant effect on AChE activity at sublethal concentrations (100 µM), suggesting AChE inhibition is not the cause of the observed behavioral effects. None of the test formulations had a significant effect on GST activity at 100 μM, suggesting that the observed behavioral phenotypes are not correlated with oxidative stress.

Previous work found that D. japonica exposed to 32 mg/L (approximately 189 µM) glyphosate ae for 3 or 5 days had altered GST expression and activity (Zhang et al, 2020), though the directionality of the effects on GST activity differed depending on the day tested. On day 3, increased GST expression and activity was observed compared to controls, whereas on day 5 GST expression was indistinguishable from controls and decreased activity was observed. In agreement with our findings, 96 h exposure in developing zebrafish to Roundup^® Ultramax did not affect oxidative stress biomarkers even at concentrations that induced lethality and developmental effects (Lanzarin et al, 2019). Effects on gene expression of several oxidative stress-related markers was upregulated in Caenorhabditis elegans following 24 h glyphosate exposure; however, only at concentrations above the lethality LOEL (García-Espiñeira et al, 2018). Thus, although oxidative stress has been suggested to be connected to the potential neurotoxic effects of GBHs (Costas-Ferreira et al, 2022), these effects may only be relevant at high concentrations which non-specifically induce systemic toxicity/lethality.

The observation that the GBHs were more toxic than pure glyphosate suggests that the additional components of the GBHs either enhance the toxicity of glyphosate or are toxic on their own. Recent work has found that several GBHs (including RC), but not glyphosate up to 10 mM, were cytotoxic and genotoxic in human cell culture (Smith-Roe et al, 2023). This agrees with a similar study comparing the cytotoxicity of 14 different GBHs and pure glyphosate in human embryonic kidney cells. This work found that the cytotoxicity of the GBHs (1/LC₅₀ [concentration that induced 50% lethality]) could be up to 358 times greater than pure glyphosate, though large variability was also seen across the different formulations (Defarge et al, 2018).

As there are over 750 different GBHs sold in the U.S. (Henderson et al, 2010), with varying—and largely unknown—formulations, it can be difficult to dissect which specific co-formulants may drive GBH toxicity. In this study, the observed toxicity of RC in planarians may be partially attributed to the other active ingredient, diquat dibromide, a non-selective herbicide and desiccant. We found that diquat had strong effects on planarian body shapes and locomotion, especially in adult planarians. In human cell genotoxicity tests, diquat dibromide alone was cytotoxic, genotoxic, and clastogenic, likely due to its effects on oxidative stress. However, RC was not found to be genotoxic (Smith-Roe et al, 2023). In contrast, the U.S. EPA has classified diquat dibromide as “not likely to be carcinogenic” due to lack of effects seen in vivo animal studies (USEPA United States Environmental Protection Agency, 1995). Together, these suggest that diquat dibromide may not be genotoxic in vivo or at the low concentrations (0.73%) used in RC. Diquat dibromide has been shown to be lethal to a variety of aquatic organisms with LC₅₀ values ranging from ∼1 to 300 mg/L (Wilson and Wu, 2012). We estimate that the concentration of RC that was lethal to adult planarians (316 µM) contained ∼2.9 mg/L diquat dibromide. Diquat alone only induced statistically significant lethality (29%) starting at 9.2 mg/L in adult planarians at day 12. However, sublethal concentrations of diquat induced a large range of locomotor and morphological defects that differed slightly from the toxicological profile of RC, though the LOELs were equivalent (RC at 31.6 µM glyphosate ae contains 0.29 mg/L diquat). Future work could explore whether the binary mixture of glyphosate and diquat dibromide could recapitulate the toxicity profile of RC or whether interactions with other adjuvants may also play a role, to better understand the toxicity mechanisms. For practical purposes, it is important to keep in mind that these concentrations are all well above the maximum expected environmental concentration (0.0031 mg/L) (Wilson and Wu, 2012).

RR contains a different active ingredient beside glyphosate: “pelargonic acid and related fatty acids”. Pelargonic acid, also called nonanoic acid, is a naturally occurring nine-carbon fatty acid that is used as an herbicide. Pelargonic acid is considered to have little to no risk for mammalian toxicity and is even approved as a food additive in the U.S. While mild toxicity has been observed with pelargonic acid in zebrafish (Techer et al, 2015), we observed no significant effects with up to 282 mg/L pelargonic acid alone. Thus, these data suggest that the toxicity of RR cannot be explained by the toxicity of its active ingredients (glyphosate and pelargonic acid) when tested individually. However, it is possible that synergistic effects of the different components enhance their toxicity.

GBHs also contain other additives such as surfactants or contaminants such as heavy metals, which have been suggested to have their own toxicity, especially to aquatic life (Defarge et al, 2018). For example, the surfactant polyoxyethylene 15 (POE 15) tallow amine and a GBH containing it were found to be 10- to 40-times more toxic than pure glyphosate to a variety of aquatic invertebrate species (Folmar et al, 1979), with the toxicity of the surfactant alone sharing many similarities to that of the GBH. Similar increases in toxicity of POE 15 over glyphosate have also been found in regulatory animal tests (Mesnage et al, 2019) and in human cell culture studies (Defarge et al, 2018). GBHs containing POEAs were also found to be correlated with more severe symptoms of acute toxicity in human poisoning cases than GBHs without POEAs (Langrand et al, 2020). The identity and concentrations of these other adjuvants is largely unknown because commercial formulations are proprietary. This makes it difficult to parse out the exact causes of toxicity in these formulations as the individual components cannot be directly tested. Thus, while we can say with confidence that pure glyphosate has significantly lower toxicity to D. japonica than the two tested GBHs, more comparative studies evaluating the toxicity of glyphosate and various GBHs are necessary to better understand this toxidrome.

Both GBH formulations had a lethality LOEL of 316 µM (53.4 mg/L) and behavioral LOELs of 31.6 µM (5.34 mg/L) and 100 µM (16.9 mg/L, RC and RR, respectively) in adult D. japonica planarians. These results agree with previously published toxicity values for 30% glyphosate isopropyl amine salt in D. japonica, which found a 96 h LC₅₀ of 128 mg/L glyphosate ae, though exposure conditions differed (bulk exposure with daily renewal). A related planarian species, G. tigrina, was found to be more sensitive to glyphosate/GBH exposure with a 48 h LC₅₀ of 36 mg/L (Córdova López et al, 2019), though behavioral and regeneration effects following 96 h exposure were seen at similar concentrations (starting at 3.75 mg/L). Notably, a different GBH formulation (Roundup^® Original), which does not contain any additional active ingredients but lists ethoxylated tallowamines as an inert ingredient, was used in the G. tigrina study. Here, we observed increased sensitivity of both G. tigrina and D. dorotocephala planarians compared to D. japonica to the same GBHs. This emphasizes that different planarian species can have different sensitivities to chemicals, as also reported in (Van Huizen et al, 2017; Ireland et al, 2020). Thus, care needs to be taken when making comparisons across species—even with the same chemical.

While this and previous studies have shown that different GBHs pose a toxic hazard to freshwater planarians, to understand whether this is of ecotoxicological concern requires consideration of environmental concentrations. Glyphosate is highly absorbed in the soil, where it is degraded by microorganisms (Matozzo et al, 2020). The rate of degradation depends on several environmental factors (Lacroix and Kurrasch, 2023), with different reports citing 50% dissipation rates ranging from 1.2—197.3 days (Saunders and Pezeshki, 2015). Because of this rapid degradation, glyphosate is generally found at low concentrations in the terrestrial and aquatic environment (Solomon, 2016; Matozzo et al, 2020). Reviews of studies from various aquatic ecosystems across the world reported water levels generally on the order of a few µg/L (Matozzo et al, 2020; Saunders and Pezeshki, 2015) but as high as a few mg/L in several studies (Edwards et al, 1980; Tzaskos et al, 2012; Avigliano and Schenone, 2015; Zhang et al, 2016). Thus, although overall environmental glyphosate levels appear to be low, because of differences in agricultural practices, soil properties and rainfall, there is the potential for contamination hot spots that may be of toxicological concern (Maggi et al, 2020). Therefore, from an ecotoxicological point of view, there appears to be a sufficiently protective margin of safety for GBH toxicity to several planarian species under normal environmental conditions. From a human neurotoxicity/developmental neurotoxicity standpoint, comparison to the maximum contaminant level for glyphosate in the U.S. (0.7 mg/L) may indicate potential risk, given that the goal is to protect the most vulnerable populations. Thus, generally a factor of 100 is used when extrapolating from animal models to humans (Konietzka et al, 2014). Moreover, exposure levels would likely be even greater for occupational workers that directly handle GBHs.

Across all endpoints, regenerating planarians showed less sensitivity to the compounds than adult planarians. For RC, most of the same endpoints were affected at both developmental stages, but with effects occurring at higher concentrations in the regenerating planarians. The toxicological profile of regenerating planarians exposed to 316 µM RC suggests overt systemic toxicity as these planarians had abnormal morphology and were practically immobile. Thus, although this concentration was not significantly lethal in regenerating planarians, it is likely that lethality would have manifested with longer exposure. On the other hand, 316 µM RR caused significant lethality in both developmental stages, but with greater magnitude in adult planarians, while mild sublethal effects were only observed in adult planarians.

While known developmentally neurotoxic compounds, such as the organophosphorus pesticide chlorpyrifos, show increased potency in regenerating planarians compared to adult planarians (Zhang et al, 2019), here, none of the GBHs showed developmental selectivity when comparing nominal concentrations. Notably, it is possible that the in vivo concentrations differ between the two planarian developmental stages due to metabolic or pharmacokinetic differences between the two developmental stages, as previously discussed (Ireland et al, 2022). This could explain the apparent differences in sensitivity when only comparing nominal concentrations.

When compared to chronic exposure studies with other popular non-mammalian organismal models, such as developing zebrafish, nematodes, or frogs, we found that the general trends were conserved. For example, studies in developing zebrafish (de Brito Rodrigues et al, 2019), nematodes (Jacques et al, 2019), or frogs (Turhan et al, 2020; Howe et al, 2004) that had direct comparisons found that GBHs (of various formulations) were more toxic than pure glyphosate. Moreover, similar to the results reported here in planarians, non-lethality effects with pure glyphosate were often only seen at concentrations that also induced significant lethality (Sulukan et al, 2017; García-Espiñeira et al, 2018; Gaur and Bhargava, 2019; Lu et al, 2022). A few sublethal developmental effects, such as malformations (Zhang et al, 2017) and premature hatching (Fiorino et al, 2018), have been observed in some developing zebrafish studies, though in other studies these same phenotypes were either not present or only at lethal concentrations (Sulukan et al, 2017; de Brito Rodrigues et al, 2019; Gaur and Bhargava, 2019). Sublethal behavioral effects on head thrashing have also been observed after 24 h glyphosate exposure in nematodes (Wang et al, 2017).

Our data with a day 12 lethality LOEL of 169 mg/L ae glyphosate show good agreement with a previous study on D. japonica that reported a day 4 LC₅₀ of 128 mg/L ae when using the glyphosate isopropyl amine salt (Zhang et al, 2023). However, when compared to other models, we found that intraspecies potency of pure glyphosate differed several orders of magnitude across published studies. For example, LOELs for lethality after 96 h varied from 0.05 to 400 mg/L ae in developing zebrafish (Sulukan et al, 2017; Zhang et al, 2017; de Brito Rodrigues et al, 2019; Gaur and Bhargava, 2019) and from 18.5 to 11840 mg/L ae in nematodes (Wang et al, 2017; Jacques et al, 2019). Differences in experimental methods and analysis can explain some of these discrepancies. These differences are even greater if studies with various GBHs are also considered. Given these vast differences in results, more studies into GBH toxicity are warranted, especially comparative studies testing many GBHs in parallel in the same system, to minimize confounding factors.

To perform large-scale comparative studies of many GBHs (and their individual components), inexpensive HTS methods that balance sensitivity and specificity are indispensable. While sensitivity is important to ensure potential hazards are identified - especially for a first-tier rapid screening model - if a system is overly sensitive, it cannot fulfill its purpose of decreasing the number of compounds that are being moved to the next tier of testing (Plunkett et al, 2010). As shown here, HTS in D. japonica was able to recapitulate the major findings observed with other common test systems, including human cell culture, and with comparable sensitivity. This suggests that planarians are a relevant and suitable system for evaluating glyphosate/GBH toxicity. The high-throughput capabilities of this system (Zhang et al, 2019) allow for large-scale comparative studies of a class of compounds. For example, planarian HTS and behavioral barcoding has been used to compare the phenotypic profiles of organophosphorus pesticides and connect specific behavioral effects with putative mechanisms (Ireland et al, 2022). In addition to their use as a biomonitor for ecotoxicity (Wu and Li, 2018), planarians are relevant for studying potential human health effects because of the complexity and highly conserved nature of the planarian nervous system (Ireland and Collins, 2023). Finally, planarians provide the unique capability to study both adult and developing brains in parallel using the same HTS assays, which allows for the distinction of neurotoxic from developmental neurotoxic effects. Together, these strengths make them a powerful model for studying GBH toxicity.